Exploration of the Usage of Color Terms by Color-blind Participants in Online Discussion Platforms

Ella Rabinovich, Boaz Carmeli

Abstract

Prominent questions about the role of sensory vs. linguistic input in the way we acquire and use language have been extensively studied in the psycholinguistic literature. However, the relative effect of various factors in a person's overall experience on their linguistic system remains unclear. We study this question by making a step forward towards a better understanding of the conceptual perception of colors by color-blind individuals, as reflected in their spontaneous linguistic productions. Using a novel and carefully curated dataset, we show that red-green color-blind speakers use the "red" and "green" color terms in less predictable contexts, and in linguistic environments evoking mental image to a lower extent, when compared to their normal-sighted counterparts. These findings shed some new and interesting light on the role of sensory experience on our linguistic system.

Introduction

Does our physical experience of the world fundamentally shape the way we speak? This research investigates a classic debate in psycholinguistics: the role of sensory versus linguistic input in conceptual development. According to embodied cognition theory, our sensory experiences—seeing, touching, and hearing—form the bedrock of our conceptual knowledge. If this theory holds, individuals with sensory deficiencies, such as color blindness (CB), should exhibit a distinct "linguistic fingerprint" when using terms related to their impaired perception.

While previous studies have explored language acquisition in blind children, there has been a lack of large-scale data regarding how subtle sensory deficiencies manifest in the spontaneous, everyday language of adults. This study addresses this gap by analyzing millions of Reddit posts to determine if the perceptual experience of color-blind individuals leaves a detectable trace in their digital footprint.

Methodology

The researchers employed a computational approach combining natural language processing (NLP) with psycholinguistic metrics to analyze a newly curated dataset from Reddit.

Datasets

  • CB Dataset: 10.1 million sentences from 2,523 authors who self-identified via flairs as having red-green color blindness (deutan or protan).
  • NS Dataset: 45.7 million sentences from 14,055 normal-sighted (NS) authors, balanced to match the CB group’s gender ratio and topical interests.

Analytical Framework

  1. Agnostic Production Filtering: To ensure the study captured natural usage rather than discussions about color blindness, the authors excluded sentences containing meta-words like "colorblind" or "vision" and removed subreddits dedicated to the condition.
  2. Imageability Modeling: Imageability measures how easily a word evokes a mental image (e.g., "apple" vs. "purity"). The authors trained a beta regression model on fasttext word embeddings to predict imageability scores for over 20,000 nouns.
  3. Contextual Word Predictability (CWP): Using the BERT masked language model, the researchers calculated how "predictable" a color term was in a given sentence. Since BERT is trained on text primarily produced by normal-sighted individuals, a higher "surprise" rank in CB text indicates a divergence from standard usage patterns.
  4. Control Sets: To isolate the effect of color blindness, the researchers compared color terms against a control set of matched adjectives (e.g., "hot," "flat") and a random split of normal-sighted users.

Key Results

The analysis revealed that the sensory deficiency of red-green color blindness significantly impacts the linguistic application of the terms "red" and "green."

The Imageability Gap

Color-blind speakers were significantly less likely to use "red" and "green" to describe highly imageable or vivid objects. For the term "red," the mean imageability score of modified nouns was 0.744 for CB users versus 0.768 for NS users ($p < .01$). A similar significant gap was observed for "green," suggesting that without the grounding of sensory experience, CB individuals use these terms in more abstract or less visually dependent contexts.

Predictability Divergence

BERT found the usage of "red" and "green" by CB authors to be less predictable than usage by NS authors. The mean rank assigned by BERT for "red" was 109.1 in CB text vs. 94.8 in NS text. For "green," the ranks were 163.6 (CB) vs. 139.3 (NS). Lower ranks indicate higher predictability; thus, CB usage is measurably less "standard."

Sensory Specificity

Crucially, these differences were specific to the impaired colors. No significant differences in frequency or predictability were found for colors like "blue," "white," or "yellow," nor for the ten control adjectives. This provides strong evidence that the linguistic shifts are a direct result of sensory grounding rather than general differences in writing style.

Discussion and Implications

An interesting finding was the compensatory usage of the color "brown." CB users used "brown" with higher imageability nouns than sighted users did, suggesting that "brown" may act as a linguistic catch-all for colors that fall within their range of perceptual difficulty.

These findings have significant implications for the development of Large Language Models (LLMs). Because LLMs learn exclusively from text without sensory input, they may lack the grounded nuances of human language. This research suggests that bridging the "sensory gap" may be a necessary step for AI to achieve truly human-like linguistic competence.

Conclusion

By analyzing spontaneous digital communication, this study demonstrates that our physical experience of the world fundamentally shapes our linguistic choices. The divergence in how color-blind individuals use color terms highlights the vital role of sensory grounding in the human linguistic system.

Summary updated

Abstract

Prominent questions about the role of sensory vs. linguistic input in the way we acquire and use language have been extensively studied in the psycholinguistic literature. However, the relative effect of various factors in a person's overall experience on their linguistic system remains unclear. We study this question by making a step forward towards a better understanding of the conceptual perception of colors by color-blind individuals, as reflected in their spontaneous linguistic productions. Using a novel and carefully curated dataset, we show that red-green color-blind speakers use the "red" and "green" color terms in less predictable contexts, and in linguistic environments evoking mental image to a lower extent, when compared to their normal-sighted counterparts. These findings shed some new and interesting light on the role of sensory experience on our linguistic system.
BibTeX
@inproceedings{rabinovich2022exploration,
  author = {Ella Rabinovich and Boaz Carmeli},
  title = {{Exploration of the Usage of Color Terms by Color-blind Participants in Online Discussion Platforms}},
  booktitle = {Conference on Empirical Methods in Natural Language Processing},
  year = {2022},
  url = {https://openreview.net/forum?id=08EHRPW2Qv},
}
Figure